Proc. R. Soc. B (2007) 274, 1023–1028 doi:10.1098/rspb.2006.0338 Published online 6 February 2007

Experimental simulations about the effects of and fragmentation on populations facing environmental warming Camilo Mora1,2,*, Rebekka Metzger1, Audrey Rollo1 and Ransom A. Myers2 1Leigh Marine Laboratory, University of Auckland, PO Box 349, Warkworth 1241, New Zealand 2Department of Biological Sciences, Dalhousie University, Nova Scotia B3H 4J1, Populations of many are dramatically declining worldwide, but the causal mechanism remains debated among different -related threats. Coping with this uncertainty is critical to several issues about the conservation and future of , but remains challenging due to difficulties associated with the experimental manipulation and/or isolation of the effects of such threats under field conditions. Using controlled microcosm populations, we quantified the individual and combined effects of environmental warming, overexploitation and habitat fragmentation on population persistence. Individually, each of these threats produced similar and significant population declines, which were accelerated to different degrees depending upon particular interactions. The interaction between habitat fragmentation and harvesting generated an additive decline in . However, both of these threats reduced population resistance causing synergistic declines in populations also facing environmental warming. Declines in population size were up to 50 times faster when all threats acted together. These results indicate that species may be facing risks of higher than those anticipated from single threat analyses and suggest that all threats should be mitigated simultaneously, if current biodiversity declines are to be reversed. Keywords: population persistence; extinction risk; human-caused threats; ; overfishing; habitat loss

1. INTRODUCTION generation of effective conservation strategies Increase in the loss of species due to human-related threats (Schiermeier 2003; Worm & Myers 2004; Grigg et al. is a major concern in modern and conservation 2005) and quantification of extinction risks (Myers 1995; (Myers 1995; Pimm et al. 1996; Chapin et al. 2000; Sala et al. 2000; Novacek & Cleland 2001). Pimm & Rave 2000; Sala et al. 2000; Novacek & Cleland The potential mechanisms by which these threats are 2001; Jenkins 2003). The decline of wild populations, likely to affect species are relatively well acknowledged. which leads to species and consequently to Habitat fragmentation, for instance, increases the distance reductions in overall biodiversity, has been mainly between populated , which, in turn, reduces attributed to the effects of habitat fragmentation, over- population connectivity and replenishment through exploitation and global warming (see reviews by Myers immigration (Hanski 1998; Debinski & Holt 2000). The 1995; Chapin et al. 2000; Sala et al. 2000; Novacek & inflow of immigrants directly influences population Cleland 2001; Jenkins 2003; Parmesan & Yohe 2003). dynamics and by introducing new genotypes it increases The specific effects and interactions of these threats resistance and the speed of recovery (i.e. resilience) remain, however, controversial because all threats do following perturbations. The inflow of immigrants also provide rational mechanisms to explain population reduces the effects of depletion, expands the declines; threats usually occur simultaneously and as yet geographical ranges of species, sustains populations that there has been a lack of robust studies fully isolating their could not be maintained through self- and individual and combined effects (e.g. Myers 1995; affects the probability of stochastic local extinctions Novacek & Cleland 2001; Jenkins 2003; Schiermeier (Hanski 1998; Debinski & Holt 2000; Cloberg et al. 2003; Aronson et al. 2004; Buckley & Roughgarden 2004; 2001; Mora & Sale 2002). In contrast, overexploitation Worm & Myers 2004; Grigg et al. 2005). Such uncertainty removes members from the stock population beyond is regarded as one of the most important challenges in natural levels of replenishment causing populations to modern ecology (Myers 1995; Chapin et al. 2000; decline and indirectly reducing and the Sala et al. 2000; Novacek & Cleland 2001) and it is a ability to adapt to other threats (Botsford et al. 1997; major limitation in the construction of future projections Jackson et al. 2001; Myers & Worn 2002; Pauly et al. 2002, on biodiversity change (Myers 1995; Sala et al. 2000), 2003). Likewise, global warming is likely to reduce population size by causing mortality of non-resistant * Author for correspondence ([email protected]). genotypes or by upsetting thermo-sensitive processes, Electronic supplementary material is available at http://dx.doi.org/10. such as reproduction (O’Brien et al. 2000; Mora & Ospina 1098/rspb.2006.0338 or via http://www.journals.royalsoc.ac.uk. 2001, 2002; Stenseth et al. 2002; Walther et al. 2002;

Received 29 November 2006 1023 This journal is q 2007 The Royal Society Accepted 9 January 2007 1024 C. Mora et al. Human threats and extinction risk

Parmesan & Yohe 2003). Alternatively, population density the individual and combined effects of habitat fragmen- could be reduced to balance the higher energetic demands of tation, overexploitation and environmental warming on maintaining metabolism in warmed environments (Poertner population persistence. et al. 2001; Allen et al. 2002). Warming is also likely to displace optimum habitats, which may or may not be colonized depending upon species’ capabilities to migrate 2. MATERIAL AND METHODS (Warren et al. 2001). In the marine realm, warming is also (a) Study species likely to affect other environmental variables, such as rainfall We used the rotifer species, Brachionus plicatilis, as the subject and salinity, marine currents and , all of which for this study. This species was chosen because it is easy to may affect populations in different ways and certainly culture in captivity, its populations can stabilize at constant complicates identifying the mechanism that causes popu- conditions (King 1967) and without expensive equipment lations to change under warming regimes. Unfortunately, cultures remain free of contamination. Contamination can be while rational explanations exist for the mechanisms by a problem with smaller organisms and may introduce which habitat fragmentation, overexploitation and warm- confounding factors. This species is also highly prolific and ing affect biodiversity, identifying their actual effects in can alternate reproductive strategies, which maximize its natural conditions has been challenging, because field recovery from reductions in population size and enhance its experiments (to quantify and isolate their effects) are levels of to selective forces (e.g. Yoshida et al. constrained by the spatial scales at which these threats 2003). This suggests that the population responses of such operate and because, in most cases, these threats are organisms, to the factors analysed, are a very conservative occurring simultaneously. measure of the potential effect that these factors may have on A promising approach to reveal the causal link less resilient species. between environmental threats and biodiversity change is the use of microcosm experiments. Microcosms (b) Experimental populations reduce complexity allowing the isolation and We first generated a stock supply of the rotifer species. The test of specific factors (Huston 1999; Jessup et al. 2004). stock was kept in three 20 l containers, which were placed in a They also minimize confounding factors that plague room at 258C and 12 h light cycle. The stock was fed observational studies, allow a high degree of experi- approximately every 12 h with a constant supply of approxi- mental control, replication and accuracy, and using mately 230 000 cells of Nannochloropsis sp. (Instant Algae) per organisms with short generations, they provide relatively millilitre of stock. This led to a constant population size of 60 fast results even about processes that could be rotifers per millilitre after two weeks, which was maintained impossible to measure in during the span of a throughout the experiment. Four weeks after the stock was human (Jessup et al. 2004). Microcosms have been set up, two of the stock containers were mixed and used to fill useful in the understanding of many functional relation- 300 transparent 50 ml centrifuge vials, which became the ships that include environmental variables, fluctuating experimental microcosm populations. Throughout the environments, ecological interactions, life-history traits, experiment, microcosms and the remaining stock container , biodiversity, ecosystem function- were fed with the same food ratio and kept under the same ing and (e.g. Gause 1934; Hairston et al. light cycle. Vials and stock containers were constantly aerated 1968; Davies et al. 1998; McGrady-Steed et al. 1998; and approximately 50% of their water was replaced every Petchey et al. 1999; Kassen et al. 2000; Fukami & Morin other day. To minimize the risk of contamination, the water 2003; Fryxell et al. 2005; Kussell & Leibler 2005; Smith used was filtered with a 1 mm filter and irradiated with et al. 2005, reviewed by Jessup et al. 2004). On the UV light. downside, however, microcosms minimize ecosystem complexity and the multidimensionality of natural (c) Experimental design conditions. This characteristic has curiously been argued Our factorial experiment consisted of 300 replicate micro- as the strength (Drenner & Mazumber 1999; Huston cosm populations, initially at equilibrium, which were 1999; Jessup et al. 2004) and weakness (Carpenter exposed to the effects of warming, harvesting and reductions 1999) of microcosms and of the entire field of in input immigrants (see scheme in figure 1). We used experimental ecology (Huston 1999). However, reviews reductions in immigration as our surrogate for increases in on the topic (i.e. Huston 1999; Jessup et al. 2004) have habitat fragmentation. Our rationale is that habitat suggested that microcosms are a necessary step towards fragmentation increases isolation among populated patches, the understanding of complex ecological processes. Field thereby reducing the amount of immigrants replenishing experiments to assess factors, such as for instance the local populations (after Hanski 1998). It is important to effect of harvesting, habitat fragmentation and warming clarify that, here, we are only dealing with the effect of the at the scale that these occur are simply not feasible isolation produced by habitat fragmentation. Habitat because the number of other environmental factors and fragmentation results from habitat loss, which may reduce variety of species is so great that an ‘empirical model of the size of local populations if habitable area influences (such system) is both conceptually and logistically . This latter effect was not considered in the intractable’ (Huston 1999). The characteristics of present study to avoid increases in the complexity of the microcosms and their broad use support their import- experiment and risks of contamination associated with large ance for revealing the causal link between human threats experiments. Note that adding one more treatment with five and biodiversity decline, particularly, when other levels to this experiment would have increased our number approaches are scarce and when the current loss of of microcosms to 1500. In this experiment, we considered as biodiversity urgently calls for answers to mitigate controls the populations that were exposed to no warming, deterring factors. Here, we used this tool to quantify no harvesting and had the highest inputs of immigrants.

Proc. R. Soc. B (2007) Human threats and extinction risk C. Mora et al. 1025

warming (ûC per generation) 0 0.3 0.6

harvesting (per cent of the initial population removed per generation) 0 12.5 25 37.5 50

immigration (per cent of the initial population added per generation) 0 12.5 25 37.5 50

replicates R1 R2 R3 R4 Figure 1. Diagram of the experimental design. The experiment consisted of three treatments for warming, five treatments for harvesting and five treatments for immigration; each treatment with four replicate microcosm populations (300 microcosm populations were deployed in total). For display purposes, arrows are shown to depict one treatment for each threat.

The highest level of immigration was chosen as the control analysed species generation time averaged 2006 days for habitat fragmentation, because the opposite was to use (calculated in this study). Changes in population time density total deprivation of immigration, which is the actual effect of among the different treatments were analysed with a factorial this simulated threat. generalized linear model. All microcosm populations (i.e. vials) were submerged into larger transparent aquaria to allow better control of water temperature (i.e. 12 aquaria, each with 25 vials). Each 3. RESULTS aquarium was equipped with an electronic temperature (a) Independent effect of threats control (Autonics TZ4S), a temperature sensor (Autonics We found that independently all simulated threats caused ! PT100) and two 300 W heaters. Using this set-up, water significant population declines ( p 0.000001; figure 2; temperature was controlled with 0.18C accuracy. Each of the table 1 in the electronic supplementary material). In microcosm populations, in each of the 12 aquaria, was contrast to control populations (i.e. populations with no exposed to one of the five levels of harvesting and immigration warming, no harvesting and were replenished with K ranging from 0 to 50% removal or addition of the individuals immigrants), which changed only 1.4% per generation G in the initial population (figure 1). The maximum levels of ( 15 s.d.; figure 2; table 2 in the electronic supple- harvesting and immigration were chosen based on reports mentary material), all harvested populations declined about the levels of current exploitation and immigration to below this value reaching a maximum decline of K G wild populations. Four aquaria were maintained at a constant 13.4% per generation ( 6 s.d.) at 50% harvesting per temperature (i.e. 258C), whereas the other eight were warmed generation (figure 2; table 2 in the electronic supple- to 338C at a gradual rate of either 1 or 28C per week. Note that mentary material). The fact that a 50% population harvest standardized to the generation time of this rotifer species, per generation led to only a 13.4% decline in population size per generation indicates the fast population turnover these rates are equivalent to about 0.3 or 0.68Cper generation. These heating rates scale reasonably with in this rotifer species. In absolute terms, however, such a warming rates that long-lived organisms might experience, decline was 9.6 times faster than the change observed considering that global average temperatures may increase among non-harvested populations. Reductions in immigration showed negative effects on from 1.4 to 5.88C over the next century (Houghton et al. population size, when immigration was below 25% input 2001) and that regional variation can be more intense due to per generation, reaching a maximum population decline of localized thermal phenomena, such as El Nino (Fedorov & K16.6% per generation (G16.9 s.d.) when populations Philander 2000). The individuals used to replenish the were totally deprived from immigration (figure 2; table 2 microcosm populations (i.e. for the immigration treatment) in the electronic supplementary material). This was a rate were obtained from the stock. These individuals were added of decline 11.9 times faster than the change among directly to populations at constant temperatures or slowly replenished control populations (figure 2; table 2 in the warmed to the temperature of warming treatments to avoid electronic supplementary material). The fact that levels of mortality due to thermal shock. In total, four replicated immigration above 25% per generation did not cause microcosm populations were used for each level of harvesting, burst increases in population density most probably immigration and warming. Densities of the microcosm reflects density-dependent control of population size and populations were measured when warmed aquaria reached the stabilization of populations at carrying capacity. 338C and after eight weeks in the aquaria that were kept at Under control conditions of harvesting and immigra- constant temperature. tion, both heating rates produced similar population declines (figure 2). This is an interesting result because (d) Data analysis one would have expected larger population declines at the Change in population density for each microcosm population quicker heating rate. However, it is important to note that was calculated proportional to the initial density and the control conditions of this experiment included high standardized to generation time as (((final densityKinitial inputs of immigrants. It is therefore most probable that the density)/initial density)/number of rotifer generations). In the individuals counted in the treatments at both heating rates

Proc. R. Soc. B (2007) 1026 C. Mora et al. Human threats and extinction risk

10 warming

0 constant temperature Ð10 0.3ûC per generation Ð20

Ð30

Ð40 0.6 ûC per generation

Ð50

Ð60

Ð70 0 population change per generation (%) 10 20 30 0 10 40 20 30 40 50 immigration harvesting (per cent input per generation)

(per cent output per generation) Figure 2. Changes in population density, per generation, of populations exposed to harvesting and reductions in immigration, when facing constant and warming temperatures. The contours in the graph were based on averages of four replicated microcosms for each interaction of treatments. The raw data of this figure appear in table 2 in the electronic supplementary material. were immigrants. This ‘artefact’ has previously been Finally, the interaction between harvesting and immigra- observed in microcosm experiments and has been used tion was non-significant ( pZ0.71). Populations that were to point the existence of sink populations (i.e. populations intensely harvested and also deprived from immigration, located in areas of extreme conditions, where self- in control conditions (i.e. no warming), declined replenishment is low to none and populations are K25.7% (G7.5 s.d.) or 18.5 times faster than control maintained almost solely by immigration; e.g. Davies non-harvested and replenished populations (figure 2; et al. 1998). When populations were deprived of table 2 in the electronic supplementary material). This immigration, population size at the quick heating rate net decline was about the same as adding the indepen- declined 1.4 times faster than populations facing slower dent effect of both threats. environmental warming (figure 2; table 2 in the electronic supplementary material). Fast warming caused population (c) Effect of all threats combined declines of K17.5% (G8 s.d.) per generation or a decline The interaction among harvesting, immigration and 11.8 times faster than control non-warmed populations. warming was non-significant ( pZ0.19; table 1 in the electronic supplementary material). Although this third- (b) Pairwise effect of threats order interaction did not cause accelerated declines in The interaction between immigration and warming was population size, the net effect of all the three factors acting significant ( p!0.004; table 1 in the electronic supple- together led to the largest declines in population size mentary material). Although all populations showed observed in this experiment (figure 2). Populations that negative growths when facing warming (figure 2), were harvested and deprived from immigration and were deprivation of immigration accelerated such declines up in fast-warming environments declined K71.7% per to K50.8% (G20 s.d.) per generation under fast generation (G11.7 s.d.) or a decline 51.6 times faster warming. This decline was 36.2 times faster than the than control non-harvested and replenished populations change observed among control populations (figure 2; at constant temperatures (figure 2). table 2 in the electronic supplementary material). Harvesting also showed a significant interaction with warming ( p!0.05; table 1 in the electronic supple- 4. DISCUSSION mentary material). Population declines were accelerated Increases in the rate of harvesting of wild species, up to K51.3% (G12.9 s.d.) per generation, when destruction of their habitats and warming of their populations experienced high levels of harvesting and environment poses a major threat on their persistence fast warming. This was a decline 36.6 times faster than (Myers 1995; Chapin et al. 2000; Sala et al. 2000; the change observed among control non-harvested Jackson et al. 2001; Novacek & Cleland 2001; Jenkins populations at constant temperatures (figure 2; table 2 2003; Parmesan & Yohe 2003). The simultaneous in the electronic supplementary material). In general, occurrence of such threats has generated a major rapid warming synergized the negative effects of harvest- challenge to unravel their independent and combined ing and immigration deprivation (compare population effects through field experimentation, which, in turn, declines at slow and fast warming rates in figure 2). has generated uncertainty and strong controversies, but

Proc. R. Soc. B (2007) Human threats and extinction risk C. Mora et al. 1027 more importantly has precluded the development of these two factors caused synergistic declines in popu- earnest mitigation policies (e.g. Myers 1995; Novacek & lations also facing warming. It is probable that the Cleland 2001; Jenkins 2003; Schiermeier 2003; reductions in population size due to either harvesting or Aronson et al. 2004; Buckley & Roughgarden 2004; deprivation of immigration are accompanied with a loss of Worm & Myers 2004; Grigg et al. 2005). Using a genetic diversity, which may impair population resilience microcosm experiment (see §1 for general description to threats like warming. Importantly, there are empirical of pros and cons), we showed that independently each data showing accelerated decays of fragmented (Warren of these threats caused similar and significant popu- et al. 2001) and harvested (Finney et al. 2000; O’Brien lation decays. As we mentioned in §1, different et al. 2000) populations facing warming. This suggests mechanisms can be involved in such declines. In threats that habitat fragmentation and harvesting do reduce that cause direct mortality, such as harvesting and population resistance to threats such as warming, that warming, the natality may not balance the mortality the experimental synergies found here do occur in nature losses, thereby leading to negative population growth. and that species could be under higher risks of extinction Other mechanisms may involve the effects of drift or than those anticipated from single threat analyses. As a perhaps (i.e. reductions in per capita first experimental insight into the effects of harvesting, reproductive success at low population levels; e.g. habitat fragmentation and warming on population persist- Myers et al.1995) in small depleted populations, ence, our study suggests that all threats are equally capable inbreeding depletion on populations deprived from of causing deleterious effects and that they all have to be immigration and limited genetic diversity for adaptation simultaneously reduced, if their synergies are to be to selective threats, such as warming. Regardless of the avoided and if the loss of species is to be reversed. mechanism, it is clear that all threats do cause negative population growth, and therefore, are capable of driving WethankC.Neho,A.Cozens,M.Birch,J.Evans, populations to extinction. T. Wustenberg and B. Dobson for collaboration with the experiment. K. Gaston, B. Worm, X. Atalah, J. McPherson In regards to the effects of warming, it is important to and R. Ford provided valuable comments on the manuscript. note that our results indicate that faster warming per Funding was provided by the Leigh Marine Laboratory and generation, after removing the effect of immigration, led to the Sloan Foundation Census of Marine Life through Future faster population declines when compared with slower of Marine Animal Populations Project. warming. These results highlight the importance of generation time in enhancing adaptation to selective forces and in explaining why some species have declined REFERENCES in step with global warming while others have not (e.g. Allen, A. P., Brown, J. H. & Gillooly, J. F. 2002 Global Parmesan & Yohe 2003). These results also support the biodiversity, biochemical kinetics, and the energy-equiv- contention that species with long generation times are alence rule. Science 297, 1545–1548. (doi:10.1126/ more prone to the effects of warming while highlighting the science.1072380) need to reduce the speed of the current warming trend Aronson, R. B. et al. 2004 Causes of coral reef degradation. (Stenseth et al. 2002; Walther et al. 2002; Parmesan & Yohe Science 302, 1502–1504. (doi:10.1126/science.302.5650. 2003). The overall decline of populations facing any 1502b) warming also highlights the sensitivity of ecological systems Botsford, L. W., Castilla, J. C. & Peterson, C. H. 1997 The to increases in temperature (Poertner et al. 2001) and management of fisheries and marine . Science suggests that environmental heating itself is capable of 277, 509–515. (doi:10.1126/science.277.5325.509) causing negative effects on populations independent of Buckley, L. B. & Roughgarden, J. 2004 Effects of changes in other environmental factors that may change in relation to climate and . Nature 430.(doi:10.1038/nature warming (e.g. rainfall, currents, productivity, etc.). A 02717) Carpenter, S. R. 1999 Microcosm experiments have limited caution with regard to our results is that, given the relevance for and . Ecology environmental gradients in nature, warming can transform 80, 1085–1088. (doi:10.2307/177043) habitats either from suitable to unsuitable or vice versa, and Chapin, F. S. et al. 2000 Consequences of changing bio- so may have negative or positive effects on population diversity. Nature 405, 234–242. (doi:10.1038/35012241) density. In our study, population size declined with Cloberg, J., Dhondt, A. A. & Nichols, J. D. 2001 Dispersal. warming, suggesting that our results apply to cases where Oxford, UK: Oxford University Press. warming reduces habitat suitability. Davies, A. J., Jenkinson, L. S., Lawton, J. H., Shorrrocks, B. Quantifying the simultaneous effect of human-related & , S. 1998 Making mistakes when predicting shifts threats is one of the major challenges in modern ecology in species range in response to global warming. Nature and one of the main worries about the future of 391, 783–786. (doi:10.1038/35842) biodiversity (e.g. Myers 1995; Chapin et al. 2000; Sala Debinski, D. M. & Holt, R. D. 2000 A survey and overview of et al. 2000; Novacek & Cleland 2001). Our study provides habitat fragmentation experiments. Conserv. Biol. 14, an insight into the nature and magnitude of such 342–355. (doi:10.1046/j.1523-1739.2000.98081.x) Drenner, R. W. & Mazumber, A. 1999 Microcosm experi- interactions. We found that the interaction between ments have limited relevance for community and ecosys- harvesting and reductions in the input of immigration, tem ecology: comment. Ecology 80, 1081–1085. (doi:10. which results from isolation of habitats due to their 2307/177042) fragmentation, was additive (i.e. they did not show a Fedorov, A. V. & Philander, G. 2000 Is El Nin˜o changing? significant interaction). We presume that this may occur Science 288, 1997–2002. (doi:10.1126/science.288.5473. because both of these factors have a similar net effect on 1997) population size through the input and output of individ- Finney, B. P., Gregory-Eaves, I., Sweetman, J., Douglas, uals. However, the reductions in population size due to M. S. V. & Smol, J. 2000 Impacts of climatic change and

Proc. R. Soc. B (2007) 1028 C. Mora et al. Human threats and extinction risk

fishing on Pacific over the past 300 Myers, R. A. & Worn, B. 2002 Rapid worldwide depletion of years. Science 290, 795–799. (doi:10.1126/science.290. predatory fish communities. Nature 423, 280–283. 5492.795) (doi:10.1038/nature01610) Fryxell, J. M., Smith, I. M. & Lynn, D. H. 2005 Evaluation of Myers, R. A., Barrowman, N. J., Hutchings, J. A. & alternate harvesting strategies using experimental micro- Rosenberg, A. A. 1995 Population dynamics of exploited cosms. Oikos 111, 143–149. (doi:10.1111/j.0030-1299. fish stocks at low population levels. Science 269, 2005.13840.x) 1106–1108. (doi:10.1126/science.269.5227.1106) Fukami, T. & Morin, P. 2003 Productivity–biodiversity Novacek, M. & Cleland, E. E. 2001 The current biodiversity relationships depend on the history of the community extinction event: scenarios for mitigation and recovery. assembly. Nature 424, 423–426. (doi:10.1038/nature Proc. Natl Acad. Sci. USA 98, 5466–5470. (doi:10.1073/ 01785) pnas.091093698) Gause, G. F. 1934 The struggle for existence. New York, NY: O’Brien, C. M., Fox, C. J., Planque, B. & Casey, J. 2000 Dover. Climate variability and North Sea cod. Nature 404, 142. Grigg, R. W. et al. 2005 Reassessing U.S. coral reefs. Science (doi:10.1038/35004654) 308, 1740–1742. (doi:10.1126/science.308.5729.1740c) Parmesan, C. & Yohe, G. 2003 A globally coherent finger- Hairston, N. G., Allan, J. D., Colwell, R. K., Futuyma, D. J., print of climate change impacts across natural systems. Howell, J., Lubin, M. D., Mathias, J. & Vandermeer, J. H. Nature 421, 37–42. (doi:10.1038/nature01286) 1968 The relationship between and Pauly, D., Christensen, V., Gue´nette, S., Pitcher, T. J., stability: an experimental approach with protozoa and Sumaila, U. R., Walters, C. J., Watson, R. & Zeller, D. bacteria. Ecology 49, 1091–1101. (doi:10.2307/1934492) 2002 Towards in world’s fisheries. Nature Hanski, I. 1998 dysnamics. Nature 396, 418, 689–695. (doi:10.1038/nature01017) 41–49. (doi:10.1038/23876) Pauly, D., Alder, J., Bennett, E., Christensen, V., Tyedmers, Houghton, J. T., Ding, Y., Griggs, D. J., Noguer, M., van der P. & Watson, R. 2003 The future for fisheries. Science 302, Linden, P. J. & Xiaosu, D. (eds) 2001 Climate change: 1359–1361. (doi:10.1126/science.1088667) the scientific basis. Cambridge, UK: Cambridge University Petchey, O. L., McPhearson, P. T., Casey, T. M. & Morin, Press. P. J. 1999 Environmental warming alters food-web Huston, M. A. 1999 Microcosm experiments have limited structure and ecosystem function. Nature 402, 69–72. relevance for community and ecosystem ecology: synthesis (doi:10.1038/47023) of comments. Ecology 80, 1088–1089. (doi:10.2307/ Pimm, S. L. & Rave, P. 2000 Extinctions by numbers. Nature 177044) 403, 843–845. (doi:10.1038/35002708) Jackson, J. B. C. et al. 2001 Historical overfishing and the Pimm, S. L., Russell, G. J., Gittleman, J. L. & Brooks, T. M. recent collapse of coastal ecosystems. Science 293, 1996 The future of biodiversity. Science 269, 347–350. 629–638. (doi:10.1126/science.1059199) (doi:10.1126/science.269.5222.347) Jenkins, M. 2003 Prospects for biodiversity. Science 302, Poertner, H. O. et al. 2001 Climate induced temperature 1175–1177. (doi:10.1126/science.1088666) effects on growth performance, fecundity and recruitment Jessup, C. M., Kassen, R., Forde, S. E., Kerr, B., Buckling, in marine fish: developing a hypothesis for cause and effect A., Rainey, P. B. & Bohannan, B. J. M. 2004 Big questions, relationships in Atlantic cod Gadus morhua and common small worlds: microbial model systems in ecology. Trends eelpout Zoarces viviparus. Cont. Shelf Res. 21, 1975–1997. Ecol. Evol. 19, 189–197. (doi:10.1016/j..2004.01.008) (doi:10.1016/S0278-4343(01)00038-3) Kassen, R., Buckling, A., Bell, G. & Rainey, P. B. 2000 Sala, O. E. et al. 2000 scenarios for the Diversity peaks at intermediate productivity in a labora- year 2100. Science 287, 1770–1774. (doi:10.1126/science. tory microcosm. Nature 406, 508–512. (doi:10.1038/ 287.5459.1770) 35020060) Schiermeier, Q. 2003 Climate findings let fishermen off the King, C. E. 1967 Food, age, and dynamics of a laboratory hook. Nature 428,4.(doi:10.1038/428004a) population of rotifers. Ecology 48, 111–128. (doi:10.2307/ Smith, V. H., Foster, B. L., Grover, J. P., Holt, R. D., Leibold, 1933423) M. A. & DeNoyelles Jr, F. 2005 Phytoplankton species Kussell, E. & Leibler, S. 2005 Phenotypic diversity, richness scales consistently from laboratory microcosms to population growth, and informatics in fluctuating environ- the world’s oceans. Proc. Natl Acad. Sci. USA 102, ments. Science 309, 2075–2078. (doi:10.1126/science. 4393–4396. (doi:10.1073/pnas.0500094102) 1114383) Stenseth, N. C., Mysterud, A., Ottersen, G., Hurrell, J. W., McGrady-Steed, J., Harris, P. M. & Morin, P. 1998 Chan, K.-S. & Lima, M. 2002 Ecological effects of Biodiversity regulates ecosystem predictability. Nature climate fluctuations. Science 297, 1292–1296. (doi:10. 390, 162–165. 1126/science.1071281) Mora, C. & Ospina, F. 2001 Thermal tolerance and potential Walther, G., Post, E., Convey, P., Menzel, A., Parmesan, C., impact of sea warming on reef fishes from Gorgona island Beebee, T. J. C., Fromentin, J.-M., Hoegh-Guldberg, O. (eastern Pacific Ocean). Mar. Biol. 139, 765–769. (doi:10. & Bairlein, F. 2002 Ecological response to recent climate 1007/s002270100626) change. Nature 416, 389–395. (doi:10.1038/416389a) Mora, C. & Ospina, F. 2002 Experimental effects of cold, La Warren, M. S. et al. 2001 Rapid responses of British Nina temperatures in the survival of reef fishes from butterflies to opposing forces of climate and habitat Gorgona Island (eastern Pacific Ocean). Mar. Biol. 141, change. Nature 414, 65–69. (doi:10.1038/35102054) 789–793. (doi:10.1007/s00227-002-0862-1) Worm, B. & Myers, R. A. 2004 Managing fisheries in a Mora, C. & Sale, P. F. 2002 Are populations of coral reef changing climate. Nature 429, 15. (doi:10.1038/429015a) fishes open or closed? Trends Ecol. Evol. 17, 422–428. Yoshida, T., Jone, L. E., Ellner, S. P., Fussmann, G. F. & (doi:10.1016/S0169-5347(02)02584-3) Hairston, N. G. 2003 Rapid evolution drives ecological Myers, N. 1995 Environmental unknowns. Science 269, dynamics in a predator–prey interaction. Nature 424, 358–360. (doi:10.1126/science.269.5222.358) 303–306. (doi:10.1038/nature01767)

Proc. R. Soc. B (2007)